Regulation of midline expression of zebrafish sonic ... - Development

2103

Development 126, 2103-2116 (1999) Printed in Great Britain © The Company of Biologists Limited 1999 DEV4131

Intronic enhancers control expression of zebrafish sonic hedgehog in floor plate and notochord F. Müller, B.-E. Chang, S. Albert, N. Fischer, L. Tora and U. Strähle* Institut de Genetique et de Biologie Moleculaire et Cellulaire, CNRS/INSERM/ULP, BP 163, 67404 Illkirch Cedex, CU de Strasbourg, France *Author for correspondence (e-mail: [email protected])

Accepted 18 February; published on WWW 19 April 1999

SUMMARY The signalling molecule Sonic hedgehog (Shh) controls a wide range of differentiation processes during vertebrate development. Numerous studies have suggested that the absolute levels as well as correct spatial and temporal expression of shh are critical for its function. To investigate the regulation of shh expression, we have studied the mechanism controlling its spatial expression in the zebrafish. We employed an enhancer screening strategy in zebrafish embryos based on co-injection of putative enhancer sequences with a reporter construct and analysis of mosaic expression in accumulated expression maps. Enhancers were identified in intron 1 and 2 that mediate floor plate and notochord expression. These enhancers also drive notochord and floor plate expression in the mouse embryo strongly suggesting that the mechanisms controlling shh expression in the midline are conserved between zebrafish and mouse. Functional analysis in the zebrafish embryo revealed that the intronic enhancers have a complex organisation. Two

activator regions, ar-A and ar-C, were identified in intron 1 and 2, respectively, which mediate mostly notochord and floor plate expression. In contrast, another activating region, ar-B, in intron 1 drives expression in the floor plate. Deletion fine mapping of ar-C delineated three regions of 40 bp to be essential for activity. These regions do not contain binding sites for HNF3β, the winged helix transcription factor previously implicated in the regulation of shh expression, indicating the presence of novel regulatory mechanisms. A T-box transcription factorbinding site was found in a functionally important region that forms specific complexes with protein extracts from wild-type but not from notochord-deficient mutant embryos.

INTRODUCTION

(Echelard et al., 1993; Krauss et al., 1993; Johnson et al., 1994; Barth and Wilson, 1995). During gastrulation, shh is expressed in the midline mesoderm comprising notochord and prechordal plate precursors and also in the directly overlying neuroectoderm giving rise to the floor plate. Other regions of expression are the posterior limb bud and the endoderm (Echelard et al., 1993; Krauss et al., 1993; Riddle et al., 1993; Strähle et al., 1996). In addition, there is evidence that the absolute levels of secreted protein determine which cell type is induced. At high concentrations, Shh induces floor plate while at lower concentrations motor neurons and interneurons can be induced in neural tube explants in vitro (Marti et al., 1995; Roelink et al., 1995; Ericson et al., 1997). Shh binds to Patched, a multipass transmembrane protein that controls a signalling cascade in target cells (Goodrich et al., 1996; Marigo et al., 1996; Stone et al., 1996; Alcedo and Noll, 1997). Shh signalling leads to activation of Gli zinc finger transcription factors (Alexandre et al., 1996; Ruiz i Altaba, 1997; Ding et al., 1998; Matise et al., 1998), which induce expression of specific genes including that of the winged-helix

Cell-cell signalling is crucial to shape the vertebrate body. Signals from the axial mesoderm pattern the ventral neural tube (Placzek et al., 1993). Sonic hedgehog (Shh), a member of the hedgehog family (Hh) of signalling molecules induces floor plate and motor neurons, the ventral-most structures in the vertebrate neural tube (Chiang et al., 1996). In comparison to the phenotype of shh−/− mice (Chiang et al., 1996), lack of shh (named sonic-you in zebrafish) causes much milder defects in the zebrafish ventral neural tube. In particular, the floor plate forms in mutants homozygous for loss-of function alleles (Schauerte et al., 1998). Two other, members of the hedgehog (hh) gene family are expressed in domains in the zebrafish body axis overlapping with that of shh (Ekker et al., 1995; Currie and Ingham, 1996). Loss of shh function may be compensated for by these closely related members of the hh family. Numerous experiments suggest that the correct spatial and temporal expression of Shh signal is crucial for patterning

Key words: floor plate, notochord, sonic hedgehog, HNF3β, enhancer, gene regulation, zebrafish, mouse transgenic, co-injection, T-box

2104 F. Müller and others transcription factor HNF3β (Sasaki et al., 1997). There is, however, also evidence that Shh controls target gene expression by transcription factors other than Glis as in the case of COUPTFII (Krishnan et al., 1997a,b). Members of the winged helix class of proteins have been implicated in the control of shh expression. Ectopic expression of HNF3β (named Axial in the zebrafish) causes ectopic activation of shh expression (Ruiz i Altaba et al., 1995a; Sasaki and Hogan, 1994; Chang et al., 1997). Furthermore, targeted mutation of HNF3β in the mouse results in loss of shh expression in the body axis (Ang and Rossant, 1994; Weinstein et al., 1994). In agreement, the zebrafish shh promoter has been shown to contain two HNF3β (Axial)-binding sites, which are required for high HNF3β (Axial)-dependent promoter activity in a cell culture system (Chang et al., 1997). Mouse embryos homozygous for targeted mutations in the shh gene fail to form the floor plate (Chiang et al., 1996). Therefore, shh controls not only floor plate differentiation but possibly also its own expression in the floor plate, in a process that has been proposed to be mediated by HNF3β (Ruiz i Altaba et al., 1995b; Chang et al., 1997). Thus, HNF3β seems to act both downstream and upstream of shh in the developmental pathway specifying floor plate. In zebrafish, mutations identified several other genes to be required for notochord and/or floor plate development and shh expression in these tissues. Mutations in floating head (flh; zebrafish homologue of Xenopus XNot1) cause impaired development of the notochord and the floor plate (Talbot et al., 1995). Embryos carrying mutations in either cyclops or squint, which encode closely related members of the Nodal subfamily of TGFβ signalling proteins, disrupt development of the ventral neural tube and shh expression (Feldman et al., 1998; Krauss et al., 1993; Rebagliati et al., 1998; Sampath et al., 1998). Hence, signalling by Nodal-related molecules may be required for specification of the ventral neural tube and shh expression. Mutations in one-eyed pinhead, which encodes a secreted protein containing an EGF and a CFC domain (Zhang et al., 1998), indicate yet another signalling pathway required for shh expression and floor plate development in zebrafish (Schier et al., 1997; Strähle et al., 1997b). shh expression in axial mesoderm and the ventral neural tube appears to be the result of multiple converging, signalling pathways. In order to better understand the mechanism underlying expression of shh in the notochord and floor plate, we have investigated the cis-regulatory elements directing shh expression in the zebrafish embryo. We employed a co-injection approach in zebrafish embryos. This involves the detection of activating sequences by co-injection of gene fragments with a reporter construct (Müller et al., 1997). This technique exploits the rapid concatamerisation of injected DNA in fish embryos (Stuart et al., 1988; Winkler et al., 1991) and by-passes the need to generate multiple expression constructs. We identified enhancers in intron 1 and 2 that control notochord and floor plate expression. These enhancers direct midline expression in both mouse and zebrafish embryos indicating that the mechanisms of shh regulation in these tissues are conserved. Detailed analysis of the introns revealed multiple modules, two of which control predominantly notochord expression while another module controls expression in the floor plate. Detailed deletion analysis of one of the modules (activating region C, arC), which mediates predominantly notochord expression

revealed that three subregions are essential for activity. One subregion, which contains a potential T-box transcription factorbinding site, forms specific complexes with protein extracts from zebrafish embryos. Protein binding is increased in extracts from dorsalised embryos in which the shh expression domain is expanded, and decreased in extracts from flh mutants lacking the notochord. MATERIALS AND METHODS Fish stocks, embryo production The wild-type zebrafish line wtOX is derived from fish purchased from the Goldfish Bowl, Oxford, UK and has been bred for several years in the laboratory. Fish were bred and raised as described (Westerfield, 1993). The origin of the flhP3B allele has been described (Strähle et al., 1996). Plasmid construction All cloning was performed following standard procedures (Sambrook et al., 1989). Briefly, plasmids −563shh::lacZ and −2.2shh::lacZ containing the zebrafish shh promoter from −563 to +221 and approximately −2200 to +221, respectively, (numbers are relative to the proximal transcription start site) are derivatives of −563shhCAT and −2200shhCAT described previously (Chang et al., 1997). The CAT coding region in these plasmids was replaced by the KpnI-SalI fragment of pHMGFNFlacZ (kindly provided by J. Pearce), which encodes a lacZ gene with a nuclear localisation signal. Plasmids I1+I2/−2.2shh::lacZ and I2+I1/−2.2shh::lacZ were constructed by inserting the XhoI-SalI fragment (Chang et al., 1997) comprising shh downstream sequences from +221 to +5995 into the SalI site in both orientations upstream of the shh promoter in −2.2shh::lacZ. The I1+I2/−2.2shh::lacZ hybrid gene with the shh introns 1 and 2 were isolated by cleavage with NotI and KpnI. The linear fragment I1/−2.2shh::lacZ was produced by isolating the MscI-KpnI fragment from I2+I1/−2.2shh::lacZ and the linear fragment I2/shh::lacZ was generated by isolating the MscI-KpnI fragment from I1+I2/−2.2shh::lacZ. Plasmid arC/HSP68::lacZ was constructed by insertion of a PCR generated XhoI-XmaI-cleaved fragment containing the region +5168/+5631 and inserted into the phsp68PTlacZ plasmid (Kothary et al., 1989) into XhoI-XmaI sites upstream of the HSP68 promoter. The sequence of the zebrafish shh genomic gene between the transcriptional start site and the translational stop has been submitted to GenBank (Accession Number AF124382). Microinjection in zebrafish embryos −2.2shh::lacZ or HSP68::lacZ reporter genes and derivatives thereof were excised and the vector fragments were removed by agarose gel electrophoresis. Likewise, fragments for co-injection were either cut out from plasmids (intron 1 and 2) or amplified by PCR (fine mapping of ar-A, ar-B and ar-C; oligonucleotide sequences used for PCR amplification of shh intronic fragments are available on request) and then purified by agarose gel electrophoresis (Sambrook et al., 1989). DNA fragments were isolated with a Quiex DNA purification kit (Quiagen), ethanol-precipitated and redissolved in water. For injection, phenol red was added to a final concentration of 0.1%. The −2.2shh::lacZ or HSP68::lacZ hybrid gene fragments were injected at a concentration of 100 µg/ml unless otherwise stated. Co-injected fragments were provided in a 3-fold (fragments >80 bp) to 5-fold (80 bp fragments) molar excess over the reporter gene fragments. Fish eggs were dechorionated using Pronase E as described (Westerfield, 1993). Dechorionated eggs were transferred to agarcoated plastic dishes containing 10% Hank’s solution (Westerfield, 1993). and were injected with a gas-driven microinjector (Eppendorf) at the 1- to 8-cell stage.

Regulation of midline expression of zebrafish sonic hedgehog 2105 Transgenic mouse embryos Vector sequences were removed from I1+I2/−2.2shh::lacZ, arC/HSP68::lacZ and −563shh::lacZ by restriction digestion and preparative gel electrophoresis. DNA, after electroelution from agarose blocks, was further purified by sucrose gradient centrifugation (5-20% sucrose in 0.02 M Tris-HCl, 0.2 M NaCl, 2 mM EDTA, pH7.4, 0.005 mg/ml ethidium bromide) (Sambrook et al., 1989). Purified DNA was then extracted with phenol-chloroform, precipitated and redissolved in TE (10 mM Tris-HCl pH7, 0.1 mM EDTA). Fragments diluted to 4 ng/µl in TE were injected into the pronucleus of fertilized mouse eggs. Embryos were either collected at 8.5, 9.5 and 11.5 days postcoitum (d.p.c.) or grown to adulthood to establish transgenic lines. Transgenic embryos and mice were identified by PCR using DNA from proteinase-K-digested yolk sacs or mouse tails and primers specific to the zebrafish shh promoter.

This type of analysis in the zebrafish is characterised by significant mosaicism of transgene expression (Westerfield et al., 1992). We also injected the −563shh::lacZ construct into mouse embryos. At 11.5 days of embryonic development, 3 out of the 22 transgenic mice had β-gal staining in the posterior portion of the limb bud (Fig. 1). The intensity of lacZ expression as well as the size of the lacZ expression domain varied between embryos. Also ectopic expression outside of the normal shh expression domain was noted (Fig. 1B) indicating influences of the transgene integration site. These data suggest that the promoter region ( −563 to +221) contains elements that control at least partially expression in the posterior region of the limb bud in mouse embryos. However, none of these transgenic mice showed expression in the notochord and floor plate.

In situ analysis For β-gal staining, fish or mouse embryos were fixed at 4°C, 1 hour in BT-fixative (4% (w/v) sucrose, 0.12 mM CaCl2, 23 mM NaH2PO4, 77 mM Na2HPO4, 4% (w/v) paraformaldehyde) and 0.2% (w/v) glutaraldehyde. After three washes in PBS, the embryos were rinsed and incubated in staining buffer (15.4 mM Na2HPO4, 4.6 mM NaH2PO4, 150 mM NaCl, 3 mM K4Fe3(CN6)6, 3 mM K3Fe4(CN6)6) and X-gal was added to a final concentration of 0.04% (w/v). Embryos were incubated at 37°C for several hours to overnight. The pattern of expressing cells was scored by using differential interference contrast optics. The restaining of β-gal-labelled embryos with the monoclonal antibody Zn12 (Metcalfe et al., 1990; Trevarrow et al., 1990) was carried out as described previously (Strähle et al., 1993, 1997a).

Downstream sequences of the zebrafish shh gene control floor plate and notochord expression in both zebrafish and mouse embryos To search for regulatory elements required for shh expression in notochord and neural tube, we employed a co-injection

EMSA analysis Whole-cell extracts were prepared on ice from 1-day-old zebrafish embryos by rupturing the embryos in 20 mM Tris-HCl pH 7.6, 2 mM DTT, 20% glycerol, followed by extraction by freeze thawing in 20 mM Tris-HCl pH 7.6, 2 mM DTT, 400 mM KCl, 20% glycerol. The homogenate was centrifuged (10000 revs/minute, 10 minutes, 4°C) and protein concentration in supernatants was determined. 20 µg protein were incubated together with 40 bp long 32P-labelled doublestranded oligonucleotides at room temperature for 40 minutes in 10 mM Tris-HCl pH 7.6, 0.5 mM EDTA, 500 ng poly(dI-dC), 4% Ficoll and 100 mM KCl. In competition experiments, protein extracts were incubated first with cold double-stranded oligonucleotides for 10 minutes prior to the addition of the labelled probe. Complexes were separated by 6% PAGE prior to autoradiography. LiCl treatment of embryos was carried out as described (Stachel et al., 1993).

RESULTS The −563shh promoter is not sufficient for notochord and floor plate expression To test whether the zebrafish shh promoter previously characterised in a mammalian cell system (Chang et al., 1997) is sufficient to drive expression in the floor plate and notochord, we constructed lacZ reporter genes containing the shh promoter from position −563 to +221 relative to the proximal transcription start site. The purified fragment −563shh::lacZ was injected into early cleavage stage zebrafish embryos and lacZ expression was monitored at 24 hours of development (n=60 embryos). Only weak expression in a random pattern was observed (Table 1, data not shown) suggesting that the promoter region from −563 to +221 is not sufficient to direct faithful expression in the embryo.

Fig. 1. The zebrafish shh promoter from −563 to +221 directs at low frequency lacZ expression in the posterior limb bud but not in floor plate and notochord in transgenic mouse embryos. (A,B) Two examples of transient transgenic mouse embryos expressing lacZ from −563shh::lacZ. (C,D) Magnification of the forelimb buds of embryos shown in A and B. The posterior region of the forelimb buds expresses lacZ. Ectopic expression was also evident in the ventral prosencephalon, diencephalon, heart, ventral somites and the limb bud. (B,D). Embryos are oriented anterior up and dorsal to the left.

2106 F. Müller and others Table 1. Downstream enhancers of sonic hedgehog control expression in floor plate and notochord % embryos expressing lacZ* (number of embryos scored)

% β-galpositive cells in notochord and floor plate (number of cells)‡

−563shh::lacZ

13.3 (60)

0 (16)

−2.2shh::lacZ (100 µg/ml)

8.3 (60)

0 ( 34)

−2.2shh::lacZ (500 µg/ml)

26.6 (105)

13 (110)

−2.2shh::lacZ+ (I1+I2) (100+300 µg/ml)

76 (50)

72 (76)

I1+I2/−2.2shh::lacZ (200 µg/ml)

12.7 (102)

96 (54)

I2+I1/−2.2shh::lacZ (200 µg/ml)

28.7 (101)

93 (69)

I1/−2.2shh::lacZ (200 µg/ml)

45 (122)

55 (153)

I2/−2.2shh::lacZ (200 µg/ml)

42 (88)

78 (146)

Construct

*Expression in the yolk syncytial layer was excluded from the calculations. ‡Cells mean total number of cells showing β-gal activity excluding the yolk syncytial layer.

protocol in which genomic fragments were injected together with a linearised minimal promoter driving the lacZ reporter gene (Müller et al., 1997). Since transient transgene expression results in strong mosaicism in the zebrafish embryo, expression maps were plotted in which positions of lacZ-expressing cells in embryos were scored and superimposed (Westerfield et al., 1992). A larger promoter/lacZ construct (−2.2shh::lacZ) containing shh 5′ sequences up to approximately −2.2kb was used in these experiments. When this reporter construct was injected at 100 µg/ml alone, a weak and random pattern of βgal activity was observed similar to that seen upon injection of the shorter promoter construct −563shh::lacZ (Table 1; Fig. 2A). When −2.2shh::lacZ was injected at higher concentrations (500 µg/ml) expression in the ventral brain anterior to the otic vesicle, but not in the spinal cord or notochord, was frequently observed (Fig. 3A,B) suggesting that the −2.2kb promoter contains elements that weakly direct expression in the ventral brain. When a fragment comprising sequences from +221 to +5995 including intron 1 and 2 was co-injected together with −2.2shh::lacZ, (at 100 µg/ml), we found increased β-gal staining that was predominantly located in cells of the notochord and floor plate (Figs 2B, 3C). Strong expression was found in the yolk syncytial layer (YSL), irrespective of co-injected fragments (data not shown). High YSL expression was also often observed when other reporter genes were injected and is thus unlikely to reflect a specific regulatory property of the shh promoter but points rather to non-specific transgene activity of the polyploid YSL nuclei (Williams et al., 1996). To verify that the results obtained by the co-injection approach reflect regulatory activity of the downstream sequences, we constructed an expression plasmid in which the 5.8 kb fragment was cloned upstream of the shh promoter in

Fig. 2. Downstream sequences containing intron 1 and 2 of the shh gene drive expression in floor plate and notochord of zebrafish embryos. Embryos were injected with linearised and vector deleted −2.2shh::lacZ alone (A) or together with a 5.8 kb fragment comprising shh intron 1 and 2, parts of exon 1 and 3 and the entire exon 2. Embryos in C were injected with I1+I2/−2.2shh::lacZ in which the 5.8 kb fragment was cloned upstream of the −2.2shh promoter. Embryos were injected at early cleavage stages and grown until 24 hours before fixing and β-gal staining. Expressing cells were scored in injected embryos (numbers of embryos are indicated) and position of expressing cells in individual embryos were superimposed. Co-injection of reporter and enhancer fragments (B) gave the same results as injection of construct I1+I2/−2.2shh::lacZ (C) in which the enhancer fragment had been cloned upstream of the −2.2shh::lacZ reporter gene prior to injection. Red dots indicate cells that fall into the expression domain of endogenous shh. Black dots indicate ectopic expression. Ectopic expression in the yolk syncytial layer is not shown.

Regulation of midline expression of zebrafish sonic hedgehog 2107 −2.2shh::lacZ. Injection of this construct (I1+I2/− harbours intron 2 and the flanking coding sequences from 2.2shh::lacZ) into zebrafish embryos directed the same pattern position +3898 to +5995 (Fig. 5A,G). of β-gal expression as obtained when the intron-containing Both constructs resulted in a significant increase of midline fragment was co-injected (Fig. 2C); 96% lacZexpressing cells were located in the notochord and floor plate (Table 1; Fig. 2C). The same restriction of expression to notochord and floor plate was obtained when I2+I1/−2.2shh::lacZ was injected, in which the 5.8 kb fragment was inserted in the opposite orientation (Table 1 and data not shown). Transgenic mouse embryos were produced containing the fragment I1+I2/−2.2shh::lacZ to investigate whether the intronic regions would also drive expression in the notochord and floor plate of mouse embryos. One transient transgenic mouse was identified by PCR and was shown to express the transgene. Furthermore, three stable transgenic lines were established, out of which two lines expressed the transgene. Between 8 and 11.5 days, all expressing transgenic mice showed β-gal activity specifically in the notochord and floor plate albeit with a mosaic pattern (Fig. 4), while expression in the gut endoderm was restricted to a very low number of cells. When compared with previously reported expression of shh in the mouse brain (Echelard et al., 1993), the zebrafish gene does not appear to faithfully reproduce the pattern in the anterior brain, expression in the ventral diencephalon is missing. Taken together these results demonstrate first that cis-regulatory elements, which are able to drive expression in the floor plate and notochord are located in the 5.8 kb fragment that includes intron 1 and 2. Also, the regulatory mechanisms underlying shh expression in the notochord and floor plate are likely to be conserved between zebrafish and mouse. Finally, our results show that co-injection technique and transient expression analysis in zebrafish embryos provides correct spatial data, and that, together Fig. 3. Mosaic β-gal expression driven by sonic hedgehog genomic sequences. with the sampling technique, a realistic picture (A,B) β-gal expression in the head of −2.2shh::lacZ-injected embryos. (A) Embryos of the expression activity of the injected injected with 500 µg/ml −2.2shh::lacZ were double labelled by β-gal staining constructs can be obtained. (turquoise) and immunohistochemistry with the anti-HNK1/L2 antibody Zn12 Multiple elements control floor plate and notochord expression In subsequent experiments, we have focused on the regulatory elements responsible for controlling shh expression in the floor plate and notochord. To delineate the position of the regulatory element(s) in the 5.8 kb fragment more precisely, either intron 1 or intron 2 was deleted from the 5.8 kb fragment upstream of the shh promoter in −2.2shh::lacZ. The construct I1/−2.2shh::lacZ contains shh sequence from position +221 to +3898 comprising intron 1 and part of exon 1 and 2 (Fig. 5A,B) while construct I2/−2.2shh::lacZ

(brown). The ventrally located medial longitudinal fascicles (mlf) and their nuclei (nmlf) served as a reference to assess the position of β-gal-expressing cells with respect to the expression domain of endogenous shh. For comparison, see Strähle and Blader (1994). (B) Expression map collected from 23 embryos injected with −2.2shh::lacZ. Red dots indicate cells that fall into the expression domain of endogenous shh. Black dots indicate ectopic expression. −2.2shh::lacZ directs expression predominantly to the ventral midline of the brain where endogenous shh is expressed. tg; trigeminal ganglion; nmlf, nucleus of the medial longitudinal fascicle; mlf, medial longitudinal fascicle. The lacZ gene driven by the −2.2shh promoter contains a nuclear localisation signal sequence. (C) Mosaic expression of nuclear localised β-gal in the trunk of a 24 h old zebrafish embryo co-injected with −2.2shh::lacZ and the 5.8 kb downstream fragment containing intron 1 and 2. Embryo is oriented anterior to the left and dorsal up. Arrowhead with asterisk points at a nucleus of a floor plate cell whereas other arrowheads indicate two β-gal-positive nuclei in the notochord. fp, floor plate; n, notochord. Orientation of embryo is dorsal up and anterior to the left.

2108 F. Müller and others expression of lacZ, when compared to the promoter construct activator region ar-B appears to direct predominantly floor −2.2shh::lacZ (Table 1; Fig. 5B,G). However, the fragment plate expression. comprising intron 2 appeared to be more stringent in its ability ar-B and ar-C drive midline expression of a to direct expression in the midline (Table 1; Fig. 5G). heterologous promoter To define the elements more accurately, we generated a set of overlapping subfragments spanning both introns (Fig. 5A). To test whether the shh promoter is required for the activity of The fragments were tested by co-injection with −2.2shh::lacZ regions ar-B and ar-C, fragments +3099/+3597 and (100 µg/ml) for their ability to direct β-gal expression in the +5168/+5631 comprising ar-B and ar-C, respectively, were conotochord and floor plate. The fragment spanning the sequence injected with the mouse heat-shock promoter HSP68 linked to from +2033 to +3098 of intron 1 as well as the subfragment lacZ (Kothary et al., 1989). The HSP68 promoter was chosen +2033/+2371 were found to direct notochord expression as because it directs lacZ expression weakly in many tissues (Figs compared to the basal activity of the −2.2shh::lacZ fragment 6D, 7B; Müller et al., 1997). (Fig. 5F,D). The partially overlapping fragment +2351/+2848 While ar-B directed expression mostly in the floor plate showed few expressing cells in the midline not significantly higher than the control promoter construct (data not shown). These findings together with the fact that fragment +553/+2040 and subfragments +1551/+2040 and +2600/+3098 did not have an effect on β-gal activity (Fig. 5A,C) locates activator region A (ar-A) between position +2040 and +2386. While ar-A directs expression predominantly in notochord, fragment +3099/+4147 of intron 1 resulted in an increase in the number of expressing cells mostly in the floor plate and to a much lesser extent in the notochord. A similar result was obtained when the subfragment +3099/+3597 was injected (Fig. 5E). Neither the flanking fragment +2600/+3098 nor fragment +3598/+3921 showed a detectable effect. Thus activator region B (ar-B) which directs predominantly floor plate expression is located between +3099 and +3597. In intron 2, two overlapping fragments, +4880/+5374 and +5168/+5631, mediate, like ar-A of intron-1, notochord expression of the lacZ reporter gene (Fig. 5A,I). Fragment +4628/+ 5169 and subfragment +4880/+5167 did not lead to specific activation (Fig. 5H). Fragment +5346/+5631 caused only a weak enhancement of notochord and floor plate expression (data not shown). Thus, ar-C appears to be located between +5168 and +5346. In agreement, the subfragment +5168/+5374 mediated strong expression in the notochord (Fig. 7C Fig. 4. The downstream 5.8 kb fragment spanning intron 1 and 2 directs nuclear localised β-gal and data not shown). expression in notochord and ventral neural tube in transgenic mouse embryos. Mouse oocytes In summary, these results show that were injected with I1+I2/−2.2shh::lacZ and embryos were either killed at 8.5 days of multiple activator sequences in intron 1 development for β-gal staining (A) or grown to term to generate a line of stably expressing and 2 of shh control expression in the animals (B). The embryo in B is derived from an outcross of the F0 at 9.5 days of development. floor plate and notochord. While the (C,D) Transverse sections through the hindbrain and the trunk of β-gal-stained, 9.5 d.p.c. F1 two activator regions ar-A and ar-C transgenic embryo, respectively. β-gal expression is mosaic. Embryos are oriented anterior left (A) and up (B). Dorsal is up in sections (C,D). n, notochord; fp, floor plate. control mostly notochord expression,

Regulation of midline expression of zebrafish sonic hedgehog 2109 when co-injected with the HSP68::lacZ hybrid gene (Fig. 6B), ar-C mediated strong activation of the HSP68 promoter in the notochord and floor plate (Fig. 6A,C). These data show that both activator regions can work in combination with a

Fig. 5. Mapping of intronic regulatory regions by coinjection into zebrafish embryos. (A) Overview of fragments injected together with −2.2shh::lacZ. The position of fragments relative to the transcription start site are indicated. Blue bars indicate regions mediating mostly notochord expression, while red bars indicate fragments that mediate floor plate expression predominantly. Regions outlined by black bars did not show tissue-restricted expression. Examples for expression patterns of fragments with no effect are shown in C and H. The three activating regions identified arA, ar-B and ar-C are indicated by coloured boxes. Position of XhoI, MscI and SalI restriction enzyme sites used to subclone the intronic regions are shown by arrows. (B-I) Expression maps collected from injected embryos. The DNA injected is outlined below each panel. The number of injected embryos from which data were collected are given in brackets.

heterologous promoter suggesting that elements in the −2.2shh promoter are not essential for their activity. However, very few cells showed HSP68 promoter activity in the ventral brain (data not shown) suggesting that elements present in the −2.2shh

2110 F. Müller and others

A

Fig. 6. ar-B and ar-C activate the heterologous HSP68 promoter in a tissue-restricted manner. (A) Overview of constructs injected and quantitation of results. Numbers give embryos that express in the floor plate (ar-B) or notochord and floor plate (ar-C and HSP68::lacZ alone) per total number of injected embryos from one representative experiment. (B) Embryo co-injected with HSP68::lacZ and ar-B (fragment +3099/+3597) showing expression of β-gal in the floor plate (fp). (C) Embryo co-injected with HSP68::lacZ and ar-C containing fragment +5168/+5631. β-gal positive cells are present in both the floor plate (fp) and the notochord (n). (D) Embryo injected with HSP68::lacZ alone. The basal activity of this promoter is higher than that of the shh promoter resulting in a higher number of ectopically expressing cells. (E,F) Mouse oocytes were injected with ar-C/HSP68::lacZ and recovered at 8.5 days of development for β-gal staining. Whole-mount embryo is oriented anterior to the left (E), and transverse section through the trunk is shown dorsal up (F). n, notochord; fp, floor plate.

promoter are necessary for activation in this region. We also injected ar-C/HSP68::lacZ in which the fragment +5168/+5631 was cloned upstream of the HSP68 promoter. The results were the same as those obtained in the co-injection experiments (data not shown). The zebrafish shh promoter contains multiple HNF3β (Axial) sites, two of which were previously shown to be functional in a cell culture transfection assay for mediating HNF3β (Axial) activation (Chang et al., 1997). The activity of ar-C was not altered when co-injected with a shh promoter construct containing point mutations in the two HNF3β sites (Chang et al., 1997), suggesting that these two HNF3β (Axial) sites are not required for the activity of arC (data not shown).

We also injected ar-C/HSP68::lacZ into mouse oocytes. lacZ expression was observed in four injected embryos out of 44. Three embryos showed β-gal activity in the notochord and floor plate exclusively confirming functional conservation of the ar-C region (Fig. 6E,F). The ar-C is composed of multiple elements Since ar-C of intron2 showed the most dramatic effect in activating notochord expression, we characterised this region further. To narrow down the limits of ar-C, we tested a series of overlapping PCR-generated, 80 bp long fragments (Fig. 7A) by co-injection with −2.2shh::lacZ (300 µg/ml). The four fragments spanning the region between +5168 to +5367 still

Regulation of midline expression of zebrafish sonic hedgehog 2111 enriched β-gal expression in the midline (28%, 29%, 40% and 20% expressing cells were located in the notochord and floor plate, respectively, compared to 6.5% midline cells when −2.2shh::lacZ was injected alone). Fragment +5368/+5447 did not have a specific effect on the promoter activity limiting arC between positions +5168 and +5367. Since several non-

Fig. 7. Fine mapping of activating region ar-C. (A) Overview of fragments injected. Fragments giving rise to expression in the midline are indicated by blue bars. At the bottom, the position of oligonucleotides used in the mobility shift assays (Fig. 9) are outlined. Green bar indicates oligonucleotide to that specific protein complex binding was detected. (B-G) Expression maps collected from embryos injected with HSP::lacZ alone (B), or together with fragment +5168/5374 (C) or with internal deletion ∆(+5168/+5207) (D), ∆(+5208/+5247) (E), ∆(+5248/+5287) (F), or ∆(+5288/+5327) (G).

overlapping 80 bp fragments retained specific activity, it suggests that activator region C is itself composed of multiple elements. Fragments harbouring 40 bp internal deletions over the +5168 to +5327 region (Fig. 7A) were generated to test the functional relevance of the individual activator elements in the

2112 F. Müller and others

Fig. 8. Sequence of activating region ar-C. Lines above the sequence indicate the extent of deletions that impair function of ar-C. T-BS indicates position of the homology to the Brachyury monomer-binding site (a comparison with the palindromic consensus is shown at the bottom). A homology to a CCAAT-box is present at position +5310 in inverted orientation. (GenBank accession number, AF124382)

context of the whole ar-C. We utilised the HSP68::lacZ reporter in these experiments as this reporter gene provides higher activity than the −2.2shh::lacZ hybrid gene thereby facilitating the analysis. Deletion of the regions +5168/+5207, +5208/+5247 and +5288/+5327 significantly reduced specific activation in the midline (Fig. 7D,E,G). Removal of sequences between +5248 and +5287 did not affect the activity of the fragment and caused tissue-restricted expression similar to that obtained with the wild-type sequence (Fig. 7, compare F and C). These data support the notion that ar-C is composed of at least three modules that contribute to the activity of the region. In order to identify possible regulators of the ar-C region, we inspected the sequence for binding sites of transcription factors (Fig. 8). A homology to a monomer-binding site of the T-box transcription factor Brachyury was found at position +5180 in ar-C1 (Casey et al., 1998; Kispert and Herrmann, 1993). A CCAAT-box in inverted orientation is located in arC3 (Fig. 8). Neither HNF3β-(Axial) (Kaufmann et al., 1995; Overdier et al., 1994) nor Gli (Kinzler and Vogelstein, 1990; Sasaki et al., 1997)-binding sites were found in the functionally relevant regions.

LiCl treatment of cleavage stages dorsalises the zebrafish embryo (Stachel et al., 1993) and expands the domain of shh expression (data not shown). We thus reasoned that LiCl treatment would also enrich the factors specifically interacting with the shh control region in whole embryo protein extract. When protein extract from LiCl-treated embryos was incubated with the oligo+5168/+5207 probe the upper, slower migrating complex (uc) was increased while the lower complex (lc) remained unaltered in intensity (Fig. 9 compare lane 1 and 2, or 10 and 11). These results suggest that, in the uc complex, a factor is bound to the DNA fragment whose abundance in the extract or binding activity is increased by the LiCl-induced dorsalisation of embryos. Embryos homozygous for mutations in floating head (flh), the zebrafish homologue of the Xenopus homeodomain transcription factor XNot1, fail to form a notochord and have severe defects in floor plate formation and shh expression in trunk and tail (Talbot et al., 1995; Strähle et al., 1996). To assess whether mutations in flh would affect the pattern of complex formation on oligo+5168/+5207 EMSA assays were carried out with an extract from 24 h homozygous flh mutant

The ar-C is specifically bound by proteins from zebrafish embryos We next tested whether proteins from zebrafish embryos bind to the regions identified by deletion mapping. Extracts from 24 h old zebrafish embryos were prepared and electromobility shift analysis (EMSA) was carried out with overlapping 40 bp oligonucleotides covering the region from +5168 to +5327 (Figs 7, 9). When protein extracts were prepared from wildtype embryos and incubated with radioactively labelled double-stranded oligo+5168/+5207, two distinct complexes were detected (uc, lc, Fig. 9, lane 1). The same fragment when added to the reaction in a 100-fold excess as a cold competitor, abolished complex formation (Fig. 9, lane 3). In contrast, another fragment, oligo+5368/+5408, when added as a competitor did not influence formation of the uc and lc complexes (data not shown) suggesting that the extract proteins bind specifically to the sequence from +5168 to +5207. The overlapping oligo+5188/+5227 showed only weak binding (Fig. 9, lane 4), suggesting that the sequences responsible for strong complex formation with oligo+5168/+5207 are confined between +5168 to +5187, which also contains the homology to the Brachyury monomer-binding site. DNA fragments containing sequences between +5208/+5247, +5248/+5287, +5268/5307, +5268/+5327 and +5368/+5407 did not show specific complex formation under the conditions employed (Fig. 9, from lane 7 to 9, and data not shown).

Fig. 9. Electrophoretic mobility shift assays of ar-C. Radioactively labelled, double-stranded oligonucleotides (oligo) covering sequences +5168/+5207 (lanes 1-3), or +5188/+5227 (lanes 4-6) or +5368/5407 (lanes 7-9) were incubated with protein extract prepared either from wild-type embryos (wt) or from embryos treated with LiCl (li and 100× comp) or floating head mutants (flh). Addition of a 100-fold excess of cold oligonucleotide to the binding reaction (100× comp) abolished binding in extracts treated with LiCl. Only a weak retarded band is detectable with oligo+5188/+5227 while no specific complex is formed with oligo+5368/5407. Uc, upper complex; lc, lower complex.

Regulation of midline expression of zebrafish sonic hedgehog 2113 embryos. Complexes uc and lc did not form with extract from flh mutant embryos (Fig. 9, lane 12). When the same extracts were probed with a DR-5 retinoic acid response element (Chang et al., 1997), extract from flh mutant and wild-type embryos formed the same specific complexes (data not shown). These data together with the results obtained by using the LiClinduced embryo extracts strongly support the functional relevance for the complexes formed in vitro with oligo+5168/+5207. DISCUSSION We report here the identification of regulatory elements in the shh gene that mediate expression in the notochord and ventral neural tube. The −563shh promoter, which contains HNF3β (Axial)-binding sites (Chang et al., 1997), does not efficiently direct expression in these regions of the embryo. Our analysis suggests that elements in both introns of the shh gene are required for expression in notochord and floor plate. The −563shh promoter by itself gave rise to lacZ expression in the posterior limb bud in transgenic mouse embryos, suggesting that the promoter contains elements that may direct expression in these regions. The retinoic acid response element previously identified in the shh promoter could be involved in this regulation (Chang et al., 1997). However, only a fraction of mouse embryos harbouring −563shh::lacZ displayed limb expression and furthermore none of the transgenic mice that expressed the −2.2shh promoter fused to the intronic enhancers in notochord and ventral neural tube showed expression in the limb buds. Thus, additional elements must be required to mediate efficient expression in the limb buds. Regulation of expression in the ventral neural tube appears to be complex. The intronic enhancers as well as an element in the −2.2shh promoter appear to contribute to expression in the ventral neural tube. While the intronic enhancers mediate expression in the floor plate of the spinal cord, another element in the −2.2shh promoter appears to weakly drive expression in the ventral brain. However, the anterior-most expression of shh in the ventral diencephalon does not appear to be recapitulated by any of the transgenes suggesting that elements outside of these regions are involved in this aspect of shh expression. Moreover, neither the zebrafish nor the mouse embryos showed significant expression in endoderm, another site of shh expression (Echelard et al., 1993; Krauss et al., 1993; Strähle et al., 1996). Thus, the regions that we have studied here are able to direct efficient expression in the notochord and floor plate but not in the endoderm, limbs/fins and the anterior neural tube. Enhancer screening by co-injection and mosaic analysis The production of stable lines of transgenic zebrafish and faithful expression of transgenes in transgenic lines has been demonstrated recently (Long et al., 1997; Meng et al., 1997). However, given the lack of efficient integration, the number of transmitting lines obtained remains low per experiment. To overcome this problem, we have employed a transient expression assay that has previously been shown to provide information about the activity of tissue-specific regulatory elements (Reinhard et al., 1994; Westerfield et al., 1992). A

major drawback of transient expression in the zebrafish embryo is the mosaicism of transgene distribution. To obtain a spatial pattern of enhancer activity, positions of expressing cells were recorded from a large number of embryos resulting in the generation of expression maps (Westerfield et al., 1992; Müller et al., 1997). Fast delineation of regulatory elements was possible by exploiting the efficient concatamerisation of microinjected fragments, a phenomenon well characterised in Xenopus (Marini et al., 1988) and also described in fish embryos (Stuart et al., 1988; Winkler et al., 1991). It was previously shown that strong muscle and neural-tube-specific enhancers convey tissue-restricted expression to reporter genes when co-injected into zebrafish embryos (Müller et al., 1997). In this report, we show for the first time that this property of transient transgenic fish embryos can be exploited to identify novel regulatory elements in gene fragments. Our results show that this technique is not limited to intact and thus strong enhancer regions but can be utilised to dissect elements to a very high resolution. By co-injecting sequentially shorter fragments, we were able to identify enhancers in large fragments in a significantly shorter period of time than would have been required for establishing stable lines of transgenics. Multiple modules with overlapping functions are present in the intronic regions Both notochord and floor plate derive from intermingled cells in the embryonic shield, the zebrafish equivalent of the amphibian organizer (Woo and Fraser, 1995). In avian embryos, notochord and floor plate cells were even shown to be descendants of the same precursor cells in the node (Catala et al., 1996). Although there is no direct evidence for a common precursor in the zebrafish, the notochord and floor plate do share expression of a number of genes (Hatta, 1992; Krauss et al., 1993; Strähle et al., 1993; Yan et al., 1995), suggesting that these two cell types are also very closely related in fish. However, despite shared gene expression in precursor cells and a common descent, floor plate and notochord differentiate into morphologically and functionally distinct structures. In line with these differences, the regulation of the shh gene is mediated by distinct elements that control predominantly notochord expression and an element that directs predominantly floor plate expression. However, the activity of both types of elements does not appear to be entirely restricted to one or the other tissue. The notochord elements ar-A and ar-C also mediate to a lesser, but still significant, degree activation in the floor plate while the converse holds true for the floor plate enhancer ar-B. This implies that some regulatory factors are employed in both tissues. This promiscuous activity of the activating regions may be a reflection of the common descent of floor plate and notochord. Notochord expression appears to be redundantly controlled by two activator regions. However, fragments that contain both intronic enhancer regions appear to direct more specific expression in notochord and floor plate. Moreover, we can not entirely rule out the possibility that certain elements may have unique functions that will only be evident in the context of the intact shh gene. The regulation of shh expression in the midline is dynamic: it is first expressed in the embryonic shield and then in the maturing notochord and midline of the neural plate. While expression is maintained in the neural tube, expression

2114 F. Müller and others is turned off with an anterior-to-posterior progression in the notochord during somitogenesis (Krauss et al., 1993). Therefore, it remains to be established whether these multiple elements are truly redundant or whether they are involved in specific temporal aspects of shh regulation. Which factors are involved? The presence of a sequence with homology to one half of the palindromic Brachyury, (no tail, ntl in zebrafish, SchulteMerker et al., 1994) binding site in ar-C1 is intriguing. The Xenopus eFGF gene is regulated by Xbra, the Xenopus homologue of Brachyury, through a response element also containing only one half of the palindrome which was previously shown by binding site selection to be the preferred recognition sequence for mouse Brachyury in vitro (Casey et al., 1998; Kispert and Hermann, 1993). It is unlikely, however, that ntl is indeed involved in regulation of shh expression: onset of shh expression in the shield is normal in ntl mutants. At later stages, ntl mutants have an expansion of floor plate, most likely as a result of a cell fate change of notochord cells to floor plate, and the widened floor plate appears to express shh at normal levels (Halpern et al., 1993, 1997; Krauss et al., 1993; Odenthal et al., 1996; Strähle et al., 1996). Other related T-box genes are expressed in the midline, however, and they may have a function in regulating shh expression (Dheen et al., 1999). Many lines of evidence suggest that HNF3β (Axial) is crucial for development of the notochord and the floor plate. Mutant mouse embryos lacking HNF3β activity lack the notochord, floor plate and also shh expression in the midline (Ang and Rossant, 1994; Weinstein et al., 1994). Overexpression of HNF3β results in ectopic activation of shh expression (Chang et al., 1997; Ruiz i Altaba et al., 1995a; Sasaki and Hogan, 1994). Furthermore, the zebrafish shh promoter contains HNF3β-binding sites, which mediate activation by HNF3β (Axial) in a cell system, suggesting that HNF3β (Axial) is a direct regulator of the shh gene (Chang et al., 1997). We did not find any sequence with homology to the HNF3β (Axial) consensus recognition site (Kaufmann et al., 1995; Overdier et al., 1994) in the functionally important regions of ar-C, suggesting that the activity of these regions depends on factors other than HNF3β. Given the low resolution of the functional analysis of ar-A and ar-B, it remains unclear whether the HNF3β (Axial)-binding site homologies found in these regions (F. M. and U. S., unpublished) are functionally relevant. The observation that both ar-B and ar-C act on the heterologous HSP68 promoter shows that the activity of these elements is not restricted to the shh promoter. ar-C appears to activate the mutant −563shh promoter with mutations in the two HNF3β sites that mediate activation by HNF3β (Axial) in a cell system. Thus, the two HNF3β-binding sites do not appear to be required for activation in our assay system in the embryo. These findings do not, however, rule out a requirement of HNF3β for promoter activity as a number of other HNF3βbinding site homologies are present in the promoter region that could potentially compensate for the mutations. Other promoter elements may substitute for the HNF3β in the more ubiquitously expressed HSP68 promoter. Previously, it has been proposed that Shh controls floor plate differentiation and floor-plate-specific gene programs via posttranscriptional activation of Gli-1, which in turn switches on transcription of HNF3β ( Lee et al., 1997; Sasaki et al., 1997).

ar-B, which mediates floor plate expression, harbours a sequence (+3260 TAGCACCCA) with significant homology to a Gli-1-binding site (Kinzler and Vogelstein, 1990; Sasaki et al., 1997). This homology suggests that ar-B may be a target of Gli-1 and that shh may regulate its own expression directly via Gli-1. The regulation of shh expression in notochord and floor plate are conserved between mouse and zebrafish While it is clear that shh is crucial for development of the floor plate in the mouse embryo (Chiang et al., 1996), the picture is less clear in zebrafish. Zebrafish embryos carrying loss-offunction alleles in sonic-you encoding the zebrafish orthologue of shh form floor plate and express floor plate markers (Schauerte et al., 1998). These observations raise the possibility that the regulatory mechanisms underlying floor plate specification in the zebrafish may be different from those in the mouse. Our finding that the zebrafish intronic enhancers also drive expression in the notochord and floor plate of the mouse argues strongly against such an interpretation. Furthermore, the ar-C sequence delineated by deletion mapping shares high sequence similarity with the floor plate enhancer SFPE2, which was recently identified in intron2 of the mouse shh gene (Epstein et al., 1999). Highest homology was scored in the subregions ar-C2 and ar-C3 shown by deletion analysis to be required for full activity of ar-C in the co-injection approach in the zebrafish. A homology to the Tbox transcription-factor-binding site that resides in ar-C1 of the zebrafish enhancer was found 22 bp downstream of the mouse region homologous to ar-C3 (F. M. and U. S. unpublished). Thus, zebrafish and mouse embryos appear to employ the same regulatory sequences to drive expression of the shh gene. The mechanism underlying floor plate development in the two organisms appears to be conserved. In view of this conservation and the floor plate deficiency and impaired shh expression in cyc and oep mutant zebrafish embryos (Hatta et al., 1991; Schier et al., 1997; Strähle et al., 1997b), it is likely that these genes, which encode Nodalrelated and EGF domain-containing signals, respectively (Rebagliati et al., 1998; Sampath et al., 1998; Zhang et al., 1998), are also required for floor plate development in higher vertebrates. Future experiments will have to show whether oep and cyc act directly on shh expression and if so whether they act through the intronic elements. We thank Drs F. Guillemot and J. Pearce for plasmids. We are indebted to Dr Marianne Lemeur, Corinne Marfing and Michael Gendron for the generation of mouse transgenics, to Omar Abdel Samad for help with the injections, to Dr Carol Fode for advice with mouse transgenics and to Drs Siew-Lan Ang and Nicholas S. Foulkes for critical reading of this manuscript. We are grateful to Odile Kwunda, Abdo Karmin, Dominique Biellmann for care of the fish. and to Drs B. and C. Thisse for running the fish facility. Furthermore, we are thankful to the photographers of the IGBMC, to Thomas Ding for sectioning, to the staff of the oligonucleotide synthesis and sequencing facilities. F. M. was supported by an EMBO fellowship and by FRM. B.-E. Chang was a recipient of a fellowship from the government of the Republic of China. This work was supported by the Institute National de la Sante et de la Recherche Medicale, the Centre National de la Recherche Scientifique, the Hopital Universitaire de Strasbourg, ARC, AFM, CNRS Projet Genome and La Ligue contre le Cancer.

Regulation of midline expression of zebrafish sonic hedgehog 2115 REFERENCES Alcedo, J. and Noll, M. (1997). Hedgehog and its patched-smoothened receptor complex: a novel signalling mechanism at the cell surface. Biol. Chem, 378, 583-590. Alexandre, C., Jacinto, A. and Ingham, P. W. (1996). Transcriptional activation of hedgehog target genes in Drosophila is mediated directly by the cubitus interruptus protein, a member of the GLI family of zinc finger DNA-binding proteins. Genes. Dev. 10, 2003-2013. Ang, S. L. and Rossant, J. (1994). HNF-3 beta is essential for node and notochord formation in mouse development. Cell 78, 561-574. Barth, A. K. and Wilson, S. W. (1995). Expression of zebrafish nk2.2 is influenced by sonic hedgehog/vertebrate hedgehog-1 and demarcates a zone of neuronal differentiation in the embryonic forebrain. Development 121, 1755-1768. Casey, E. S., O’Reilly, M.-A. J., Conlon, F. L. and Smith, J. C. (1998). The T-box transcription factor Brachyury regulates expression of eFGF through binding to a non-palindromic response element. Development 125, 38873894. Catala, M., Teillet, M.-A., De Robertis, E. M. and Le Douarin, N. M. (1996). A spinal cord fate map in the avian embryo: while regressing Hensen’s node lays down the notochord and floor plate thus joining the spinal cord lateral walls. Development 122, 2599-2610. Chang, B.-E., Fischer, N., Blader, P., Ingham, P. and Strähle, U. (1997). Axial (HNF3b) and retinoic acid receptors are regulators of the zebrafish sonic hedgehog promoter. EMBO. J. 16, 3955-3964. Chiang, C., Litingtung, Y., Lee, E., Young, K. E., Corden, J. L., Westphal, H. and Beachy, P. A. (1996). Cyclopia and defective axial patterning in mice lacking sonic hedgehog gene function. Nature 383, 407-413. Currie, P. D. and Ingham, P. W. (1996). Induction of a specific muscle cell type by a hedgehog-like protein in zebrafish. Nature 382, 452-455. Dheen, S. T., Sleptsova-Friedrich, T., Xu, Y., Clark, M., Lehrach, H., Gong, Z. and Korzh, V. (1999). Zebrafish tbx-c functions during formation of midline structures. Development (in press). Ding, Q., Motoyama, J., Gasca, S., Mo, R., Sasaki, H., Rossant, J. and Hui, C. C. (1998) Diminished Sonic hedgehog signaling and lack of floor plate differentiation in Gli2 mutant mice. Development 125, 2533-2543. Echelard, Y., Epstein, D. J., St-Jacques, B., Shen, L., Mohler, J., McMahon, J. A. and McMahon, A. P. (1993). Sonic hedgehog, a member of a family of putative signaling molecules, is implicated in the regulation of CNS polarity. Cell 75, 1417-1430. Ekker, S. C., Ungar, A. R., Greenstein, P., von Kessler, D. P., Porter, J. A., Moon, R. T. and Beachy, P. A. (1995). Patterning activities of vertebrate hedgehog proteins in the developing eye and brain. Curr. Biol. 5, 944-955. Ericson, J., Rashbass, P., Schedl, A., Brenner-Morton, S., Kawakami, A., van Heyningen, V., Jessell, T. M. and Briscoe, J. (1997). Pax6 controls progenitor cell identity and neuronal fate in response to graded Shh signaling. Cell 90, 169-180. Epstein, D. J., McMahon, A. P. and Joyner, A. L. (1999) Regionalization of sonic hedgehog transcription along the anterioposterior axis of the mouse centarl nervous system is regulated by Hnf3-dependent and -independent mechanisms. Development 126, 281-292. Feldman, B., Gates, M. A., Egan, E. S., Dougan, S. T., Rennebeck, G., Sirotkin, H. I., Schier, A. F. and Talbot, W. S. (1998). Zebrafish organizer development and germ layer formation require nodal-related signals. Nature 395, 181-185. Goodrich, L. V., Johnson, R. L., Milenkovic, L., McMahon, J. A. and Scott, M. P. (1996). Conservation of the hedgehog/patched signaling pathway from flies to mice: induction of a mouse patched gene by Hedgehog. Genes. Dev. 10, 301-312. Halpern, M. E., Ho, R. K., Walker, C. and Kimmel, C. B. (1993) Induction of muscle pioneers and floor plate is distinguished by the zebrafish no tail mutation. Cell 75, 99-111. Halpern, M. E., Hatta, K., Amacher, S. L., Talbot, W. S., Yan, Y. L., Thisse, B., Thisse, C., Postlethwait, J. H. and Kimmel, C. B. (1997). Genetic interactions in zebrafish midline development. Dev. Biol. 187, 154-170. Hatta, K., Kimmel, B. C., Ho, R. K. and Walker, C. (1991) The cyclops mutation blocks specification of the floor plate of the zebrafish central nervous system Nature. 350, 339-341. Hatta, K. (1992). Role of the floor plate in axonal patterning in the zebrafish CNS. Neuron 9, 629-642. Johnson, R. L., Laufer, L., Riddle, R. D. and Tabin, C. (1994). Ectopic expression of sonic hedgehog alters dorsal-ventral patterning of somites. Cell 79, 1165-1173.

Kaufmann, E., Müller, D. and Knöchel, W. (1995). DNA recognition site analysis of Xenopus winged helix proteins. J. Mol. Biol. 248, 239-254. Kinzler, K. W. and Vogelstein, B. (1990). The GLI gene encodes a nuclear protein which binds specific sequences in the human genome. Mol. Cell. Biol. 10, 634-642. Kispert, A. and Herrmann, B. G. (1993). The Brachyury gene encodes a novel DNA-binding protein. EMBO J. 12, 3211-3220. Kothary, R., Clapoff, S., Darling, S., Perry, M. D., Moran, L. and Rossant, J. (1989). Inducible expression of an hsp68-lacZ hybrid gene in transgenic mice. Development 105, 707-714. Krauss, S., Concordet, J.-P. and Ingham, P. W. (1993). A functionally conserved homolog of the Drosophila segment polarity gene hedgehog is expressed in tissues with polarizing activity in zebrafish embryos. Cell 75, 1431-1444. Krishnan, V., Elberg, G., Tsai, M. J. and Tsai, S. Y. (1997a). Identification of a novel sonic hedgehog response element in the chicken ovalbumin upstream promoter-transcription factor II promoter. Mol. Endocrinol. 11, 1458-1466. Krishnan, V., Pereira, F. A., Qiu, Y., Chen, C. H., Beachy, P. A., Tsai, S. Y. and Tsai, M. J. (1997b). Mediation of Sonic hedgehog-induced expression of COUP-TFII by a protein phosphatase. Science 278, 19471950. Lee, J., Platt, K. A., Censullo, P. and Ruiz i Altaba, A. (1997). Gli1 is a target of Sonic hedgehog that induces ventral neural tube development. Development 124, 2537-2552. Long, Q., Meng, A., Wang, H., Jessen, J. R., Farrell, M. J. and Lin, S. (1997). GATA-1 expression pattern can be recapitulated in living transgenic zebrafish using GFP reporter gene. Development 124, 4105-4111. Marigo, V., Davey, R. A., Zuo, Y., Cunningham, J. M. and Tabin, C. J. (1996). Biochemical evidence that Patched is the Hedgehog receptor. Nature 384, 176-179. Marini, N. J., Etkin, L. D. and Benbow, R. M. (1988). Persistence and replication of plasmid DNA micro-injected into early embryos of Xenopus laevis. Dev. Biol. 127, 421-434. Marti, E., Bumcrot, D. A., Takada, R. and McMahon, A. P. (1995). Requirement of the 19k form of sonic hedgehog for induction of distinct ventral cell types in CNS explants. Nature 375, 322-325. Matise, M. P., Epstein, D.J., Park, H.L., Platt, K. A. and Joyner, A. L. (1998) Gli2 is required for induction of floor plate and adjacent cells, but not most ventral neurons in the mouse central nervous system. Development 125, 2759-2570. Meng, A., Tang, H., Ong, B. A., Farrell, M. J. and Lin, S. (1997). Promoter analysis in living zebrafish embryos identifies a cis-acting motif required for neuronal expression of GATA-2. Proc. Natl. Acad. Sci. USA 94, 6267-6272. Metcalfe, W. K., Myers, P. Z., Trevarrow, B., Bass, M. B. and Kimmel, C. B. (1990). Primary neurons that express the l2/hnk-1 carbohydrate during early development in the zebrafish. Development 110, 491-504. Müller, F., Williams, D. W., Kobolak, J., Gauvry, L., Goldspink, G., Orban, L. and Maclean, N. (1997). Activator effect of coinjected enhancers on the muscle-specific expression of promoters in zebrafish embryos. Mol. Reprod. Dev. 47, 404-412. Odenthal, J., Haffter, P., Vogelsang, E., Brand, M., van Eeden, F. J., Furutani-Seiki, M., Granato, M., Hammerschmidt, M., Heisenberg, C. P., Jiang, Y. J., Kane, D. A., Kelsh, R. N., Mullins, M. C., Warga, R. M., Allende, M. L., Weinberg, E. S. and Nusslein-Volhard, C. (1996). Mutations affecting the formation of the notochord in the zebrafish, Danio rerio. Development 123, 103-115. Overdier, D. G., Porcella, A. and Costa, R. H. (1994). The DNA-binding specificity of the hepatocyte nuclear factor 3/forkhead domain is influenced by amino-acid residues adjacent to the recognition helix. Mol. Cell. Biol. 14, 2755-2766. Placzek, M., Jessell, T. M. and Dodd, J. (1993). Induction of floor plate differentiation by contact-dependent, homeogenetic signals. Development 117, 205-218. Rebagliati, M. R., Toyama, R., Haffter, P. and Dawid, I. B. (1998). Cyclops encodes a nodal-related factor involved in midline signalling. Proc. Natl. Acad. Sci. USA 95, 9932-9937. Reinhard, E., Nedivi, E., Wegner, J., Skene, J. H. P. and Westerfield, M. (1994). Neural selective activation and temporal regulation of a mamalian GAP-43 promoter in zebrafish. Development 120, 1767-1775. Riddle, R., Johnson, R., Laufer, E. and Tabin, C. (1993). Sonic Hedgehog mediates the polarizing activity of the ZPA. Cell 75, 1401-1416. Roelink, H., Porter, J. A., Chiang, C., Tanabe, Y., Chang, D. T., Beachy, P. A. and Jessell, T. M. (1995). Floor plate and motor neuron induction by

2116 F. Müller and others different concentrations of the amino-terminal cleavage product of sonic hedgehog autoproteolysis. Cell 81, 445-455. Ruiz i Altaba, A. (1997). Catching a Gli-mpse of Hedgehog. Cell 90, 193-6. Ruiz i Altaba, A., Placzek, M., Baldassare, M., Dodd, J. and Jessell, T. M. (1995a). Early stages of notochord and floor plate development in the chick embryo defined by normal and induced expression of HNF-3 beta. Dev. Biol. 170, 299-313. Ruiz i Altaba, A., Roelink, H. and Jessell, T. M. (1995b). Restrictions to floor plate induction by hedgehog and winged-helix genes in the neural tube of frog embryos. Mol. Cell. Neurosci. 6, 106-121. Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular Cloning, A Laboratory Manual. Cold Spring Harbor Press Sampath, K., Rubinstein, A. L., Cheng, A. M. S., Liang, J. O., Fekany, K., Solnica-Krezel, L., Korzh, V., Halpern, M. E. and Wright, C. V. E. (1998). Induction of the zebrafish ventral brain and floor plate requires Cyclops/Nodal signalling. Nature 395, 185-189. Sasaki, H. and Hogan, B. L. M. (1994). HNF-3β as a regulator of floor plate development. Cell 76, 103-115. Sasaki, H., Hui, C., Nakafuku, M. and Kondoh, H. (1997). A binding site for Gli proteins is essential for HNF-3beta floor plate enhancer activity in transgenics and can respond to Shh in vitro. Development 124, 13131322. Schauerte, H. E., van Eden, F. J. M., Fricke, C., Odenthal, J., Strähle, U. and Haffter, P. (1998). Sonic hedgehog is not required for floor plate induction in the zebrafish embryo. Development 125, 2983-2993. Schier, A. F., Neuhauss, S. C., Helde, K. A., Talbot, W. S. and Driever, W. (1997). The one-eyed pinhead gene functions in mesoderm and endoderm formation in zebrafish and interacts with no tail. Development 124, 327-342. Schulte-Merker, S., van Eeden, F. J. M., Halpern, M. E., Kimmel, C. B., and Nüsslein-Volhard, C. (1994) no tail (ntl) is the zebrafish homologue of the mouse T (Brachyury) gene. Development 120, 1009-1115. Stachel, S. E., Grunwald, D. J. and Myers, P. Z. (1993). Lithium perturbations and goosecoid expression identify adorsal specification pathway in the pregastrula zebrafish. Development 117, 1261-1274. Stone, D., Hynes, M., Armanini, M., Swanson, T. A., Gu, Q., Johnson, R. L., Scptt, M. P., Pennica, D., Goddard, A., Phillips, H., Noll, M., Hooper, J. E., de Sauvage, F. and Rosenthal, A. (1996). The tumour suppressor gene patched encodes a candidate receptor for Sonic Hedgehog. Nature 384, 129-134. Strähle, U. and Blader, P. (1994). Early neurogenesis in the zebrafish embryo. FASEB J. 8, 692-698. Strähle, U., Blader, P., Henrique, D. and Ingham, P. W. (1993). Axial, a

zebrafish gene expressed along the developing body axis, shows altered expression in cyclops mutant embryos. Genes Dev. 7, 1436-1446. Strähle, U., Blader, P. and Ingham, P. W. (1996). Expression of axial and sonic hedgehog in wildtype and midline defective zebrafish embryos. Int. J. Dev. Biol. 40, 929-940. Strähle, U., Fischer, N. and Blader, P. (1997a). Expression and regulation of a netrin homologue in the zebrafishembryo. Mech. Dev. 62, 147-160. Strähle, U., Jesuthasan, S., Blader, P., Garcia-Vilalba, P., Hatta, K. and Ingham, P. (1997b). one-eyed pinhead is required for floor plate development in the zebrafish embryo. Genes Function 1, 131-148. Stuart, G. W., McMurray, J. V. and Westerfield, M. (1988). Replication, integration and stable germ-line transmission of foreign sequences injected into early zebrafish embryos. Development 103, 403-412. Talbot, W. S., Trevarrow, B., Halpern, M. E., Melby, M. E., Farr, A. E., Postlethwait, J. H., Jowett, T., Kimmel, C. B. and Kimelman, D. (1995). Requirement for the homeobox gene floating head in zebrafish development. Nature 378, 150-157. Trevarrow, B., Marks, D. L. and Kimmel, C. B. (1990). Organization of hindbrain segments in the zebrafish embryo. Neuron 4, 669-679. Weinstein, D. C., Ruiz i Altaba, A., Chen, W. S., Hoodless, P., Prezioso, V. R., Jessell, T. M. and Darnell Jr, J. E. (1994). The winged-helix transcription factor HNF3b is required for notochord development in the mouse embryo. Cell 78, 575-588. Westerfield, M., Wegener, J., Jegalian, B. G., DeRobertis, E. M. and Püschel, A. W. (1992). Specific activation of mammalian Hox promoters in mosaic transgenic zebrafish. Genes Dev. 6, 591-598. Westerfield, M., (1993). The Zebrafish Book. University of Oregon Press, Eugene. Williams, D. W., Müller, F., Lavender, F. L., Orban, L. and Maclean, N. (1996). High transgene activity in the yolk syncytial layer affects quantitative transient expression assays in zebrafish (Danio rerio) embryos. Transgenic Res. 5, 433-442. Winkler, C., Vielkind, J. R. and Schartl, M. (1991). Transient expression of foreign DNA during embryonic and larval development of the medaka fish (Oryzias latipes). Mol. Gen. Genet. 226, 129-140. Woo, K. and Fraser, S. E. (1995). Order and coherence in the fate map of the zebrafish nervous system. Development 121, 2595-2609. Yan, Y. L., Hatta, K., Riggleman, B. and Postlethwait, J. H. (1995). Expression of a type II collagen gene in the zebrafish embryonic axis. Dev. Dyn. 203, 363-376. Zhang, J., Talbot, W. S. and Schier, A. F. (1998). Positional cloning identifies zebrafish one-eyed pinhead as a permissive EGF-related ligand required during gastrulation. Cell 92, 241-251.